Statins Confirm the Healthy Chain Signature: The M_MED Spectrum Now Has Two Anchors

The vulnerability this experiment was designed to address

After five M_MED experiments, the structural spectrum of pharmaceutical evidence chains had a critical weakness: the healthy class had only one case.
• Healthy differentiation (vaccines, M_MED2): r = +0.75, p = 0.014 -- n = 1
• Mild circularity (antidepressants, M_MED1): r = +0.41 -- n = 1
• Strong circularity (opioids, M_MED3): r = +0.14 -- n = 1
• Hybrid lock-in (Alzheimer's, M_MED5): Pearson r = +0.09 -- n = 1
• Measurement-layer dominance (schizophrenia, M_MED4): r = -0.13 -- n = 1

A critic with one question could collapse the entire spectrum: how do you know vaccines wasn't lucky?

M_MED6 was designed to answer that question directly.

Why statins

The statin cardiovascular evidence chain has three structural properties that make it the strongest possible healthy chain control:

Temporal independence. The LDL cholesterol hypothesis (Framingham Heart Study, 1961) predated statin development by decades. The theoretical claim existed before any statin was synthesized. This is the opposite of the antidepressant chain, where the monoamine hypothesis was constructed retrospectively to explain observed drug effects.

Mechanistic independence. The LDL receptor mechanism was established by Brown and Goldstein from cell biology and familial hypercholesterolemia genetics (Nobel Prize, 1985) -- not from observing statin effects. Familial hypercholesterolemia patients, whose LDL receptors are genetically non-functional, develop premature cardiovascular disease at dramatically elevated rates. This genetic validation is non-circular: it exists independently of any therapeutic intervention.

Double endpoint grounding. LDL-C as the surrogate endpoint is directly specified by two independent mechanistic levels -- the LDL hypothesis (elevated LDL causes atherosclerosis) and the receptor mechanism (LDL receptor regulates LDL clearance). Measuring LDL-C is measuring the mechanism directly, not a downstream behavioral symptom. This is structurally different from measuring HAM-D in antidepressants (symptom proxy for monoamine activity) or PANSS in schizophrenia (historically contingent symptom scale).

Pre-registration

All four hypotheses committed before analysis. Hash: c7a8423c. Timestamp: 2026-06-03T19:35:58Z. Records: github.com/vladi160/preregistrations, experiment M_MED6_v1.

Results: 4/4 confirmed

h1 CONFIRMED -- r(justifies, selects_endpoints) = +0.676, Spearman = +0.756, p = 0.030. The statin chain replicates the healthy topology of M_MED2 vaccines (r = +0.75, p = 0.014). Both cases are statistically significant at n < 15. The healthy class now has n = 2.

h2 CONFIRMED -- ldl_cholesterol_hypothesis is rank #1 in the justifies layer. The founding LDL hypothesis has out-edges to the receptor mechanism, the LDL-C surrogate endpoint, and the MACE clinical endpoint. It is the structural root of the justification chain.

h3 CONFIRMED -- clinical_guidelines_statins is rank #1 in the cites_as_support layer. The ACC/AHA and ESC guidelines collect citations from the major RCTs (4S, WOSCOPS, HPS, JUPITER), the CTT dose-response meta-analysis, the Framingham risk data, and the FDA regulatory approval -- four independent empirical sources. The founding LDL hypothesis drops in the citation layer; the empirical institutional output rises.

h4 CONFIRMED -- rct_statin_trials is rank #4 in cites_as_support. Major trial data is a top citation hub -- empirical evidence drives citations, not the founding theory.

The healthy chain structural signature

Compared side by side:

| Property | M_MED2 Vaccines | M_MED6 Statins | |---|---|---| | r(justifies, selects_endpoints) | +0.75, p=0.014 | +0.676, p=0.030 | | Top hub in justifies | adaptive_immunity_mechanism | ldl_cholesterol_hypothesis | | Top hub in cites_as_support | clinical_vaccination_guidelines | clinical_guidelines_statins | | Justifies hub = cites hub? | NO | NO | | Citation anomaly detected? | NO | NO |

Both cases show the same structural pattern: a founding theoretical claim dominates the justification layer but drops in the citation layer, while an empirical institutional output (guidelines) accumulates the citations. The founding theory and the citation hub are different nodes.

This is structurally opposite to M_MED1 (antidepressants), M_MED3 (opioids), and M_MED4 (schizophrenia), where the founding theoretical claim was rank #1 in both the justification layer and the citation layer -- the circular hub.

A new finding: citation anomalies do not appear in healthy chains

M_MED3 (opioids) detected the Porter and Jick letter as a citation anomaly -- rank #3 in citations, peripheral in justification. M_MED5 (Alzheimer's) detected Lesne 2006 as a citation anomaly -- rank #5 in citations, rank #12 (last) in justification with zero edges.

M_MED6 shows no such pattern. No node in the statin chain has high citation rank combined with zero or near-zero justification edges. Every node that accumulates citations (guidelines, FDA approval, trials) also has non-trivial justification presence.

This supports a specific falsifiable hypothesis: citation anomalies are a structural feature of single-founder circular chains, not a general property of large citation networks. If this holds across more cases, it becomes a predictive claim -- topology alone can flag citation anomaly risk without reading any content.

The updated spectrum

Seven experiments now map the full range:
• Healthy differentiation (vaccines, statins): r = +0.75, +0.68, both p < 0.05 -- n = 2
• Multi-framework convergent (epilepsy, M_MED7): r = +0.49
• Mild circularity (antidepressants): r = +0.41
• Strong circularity (opioids): r = +0.14
• Hybrid lock-in (Alzheimer's): Pearson r = +0.09
• Measurement-layer dominance (schizophrenia): r = -0.13

The gap between the healthy floor (+0.68) and mild circularity (+0.41) is 0.27. The two healthy cases are statistically significant; no other case reaches significance. The spectrum is now anchored at both ends.

What IRDME cannot say

This analysis reads graph structure, not scientific content. It cannot:
• Evaluate whether statins reduce cardiovascular mortality (that is established by the RCTs themselves)
• Determine the correct LDL treatment target for any individual patient
• Assess whether newer lipid-lowering agents (PCSK9 inhibitors, inclisiran) have the same structural profile

What it can say: the statin cardiovascular evidence chain has the structural topology of a healthy evidence chain. The founding theory and the citation hub are different nodes. The endpoint is theory-derived. No citation anomalies appear. This is structurally distinct from every circular chain tested in the M_MED series.